Scatterplots of two cloud-relevant variables (specific regions of the globe: with specific cloud regimes)

[bouweandela]

No description provided.

[bouweandela]

Recipe now available in ESMValTool: ESMValGroup/ESMValTool#3923

[bouweandela]

The current implementation to plot plot_joint_cli_ta_ref uses about 70GB of RAM and takes about 15 minutes to create. It would be nice to reduce this.

[bouweandela]

Faster version with lower memory requirements (which requires 35 seconds and 1.5 GB of RAM for the aforementioned plot) under review at ESMValGroup/ESMValTool#3992.

[bouweandela]

@hb326 recipe_ref_scatterplot.yml produces only global plots, should I add region selection? If yes, which regions? And what about the specific cloud regimes? It produces plots comparing several variables, do those represent the specific cloud regimes?

[bouweandela]

I discovered that many datasets used by this recipe are already in obs4MIPs, but when I tried to update the recipe to use them I ran into several issues:

• convert_units does not convert valid_min/valid_max SciTools/iris#6411
• Dataset problem: obs4MIPs CERES-EBAF-4-2 has invalid valid_min and valid_max attributes ESMValGroup/ESMValCore#2712
• obs4MIPs uses two different facet names for frequency ESMValGroup/ESMValCore#2711

before giving up.

@lewisjared @hb326 I suspect this may also be the case for the other recipes that use observational data. How do we proceed with moving from OBS/OBS6/native6 to obs4MIPs? Shall I add the diagnostic as-is and we ask the recipe authors to update the recipe so it works with obs4MIPs and report any problems? If there are data quality issues we contact obs4MIPs to see if they can be solved there and we try to tackle Iris/ESMValCore issues with the ESMValTool tech lead team?